Welcome to Tesla Motors Club
Discuss Tesla's Model S, Model 3, Model X, Model Y, Cybertruck, Roadster and More.
Register

17.24.28 released - improved 'silky' update from 17.22.46?

This site may earn commission on affiliate links.
So I just called my service center and told them that I still haven't seen the update. They took down my number that the car is registered under and said they will talk to a service tech and then call me back. After 15 mins I get an alert that I have an update. They have not called me back yet so it could be a coincidence or they actually pushed the update to my car. Either way I'm excited to get it and try it out.

Cue ~30,000 phone calls to Tesla service centers in the next 10 minutes.
 
  • Funny
Reactions: xyeahtony
So I just called my service center and told them that I still haven't seen the update. They took down my number that the car is registered under and said they will talk to a service tech and then call me back. After 15 mins I get an alert that I have an update. They have not called me back yet so it could be a coincidence or they actually pushed the update to my car. Either way I'm excited to get it and try it out.

Do you have HW2 (AP2)? If so, drive on freeways within close proximity to parallel running residential roads and let me know if you randomly slow down to 30-35 MPH in the middle of the highway. Also would be interested in how the braking occurs as you approach slower or stopped cars.

Thanks
 
FWIW it seems like 17.24.30 is trickling out very slowly. Doesn't feel like it'll be the mass deployment build either.
Capture.PNG


Roll-out seems even handed between AP1 & AP2.
 
Do you have HW2 (AP2)? If so, drive on freeways within close proximity to parallel running residential roads and let me know if you randomly slow down to 30-35 MPH in the middle of the highway. Also would be interested in how the braking occurs as you approach slower or stopped cars.

Thanks

The freeways by me are not in close proximity to residential roads. I'll test out the braking. I'm probably going to do a video tomorrow and I'll share it.
 
My MS still hasn't been scheduled and my MX got it via SC push. Of the few AP2 people I know that have 0.28 did not get it down the wire, they got via SC. The 0.30 came quickly down the wire as Has been confirmed to be hot fixes for problems with 0.28.

I firmly don't think .28/.30 is out in wide release for AP2s
 
I got .30 today on my MS/HW2. Had been in .17.

On the highway in Phoenix .17 was already quite good in my opinion. So, I wasn't expecting a change, but .30 has noticeably smoother handling.

On surface streets .30 is better - smoother on stops and starts. Handled well through intersections. Got a little close to the curb at one point, but overall pretty good. I don't really see the point of using it on surface streets until it gets more capabilities though.

I feel like they are making progress and it is about where I expected it to be. Very interested in the next round of updates. Also, curious as to how they are using all of the data they are now collecting to train. Would they take a regional or localized approach? Or does that even matter?
 
  • Informative
Reactions: EinSV
I got .30 this morning... and I will log some serious miles tomorrow so I can post observations on the changes from .28.... one thing I noticed already is that the image of the cars in front of you is more crisp on the display... and the lane lines are also more precise....

Also, it may be too early to tell, but it seems that my car is having to relearn one of the tight turn roads that .28 kept improving on this last week... .28 somehow learned to navigate the road... but now with .30 today, it's like the road has to be relearned... all of the these observations could be coincidence or lighting, or some other random thing like no car in front of me on the tight turn, so take with a healthy grain of salt.
 
There are many possible reasons why the same route would be handled better on one day than another. I've never seen any objective evidence, or even an explicit claim by Tesla, that our cars actually learn and change their behavior within the scope of a single release.

I really want to lose this argument! It would be wonderful to have such a feature.

Yes, to your point... I received .30 today, after 6 days and 500 miles on .28, and it is as though my car is having to relearn the same tight turn 0.28 seemed to learning by day 6.

The thing I can't account for is if my car is really learning.... .28 definitely appeared to be learning to me... too early for me to say if .30 is the same.
 
Because I'm impatient... i just went for a 40 mile drive this evening. Here is my observation of .30....
1) lane lines and the leading car are more detailed on the AP2 driver display,
2) there is continued improvement in lane confidence and staying in the center,
3) over slowing of the speed limit on a fly over transition to a different highway that was probably there on previous versions...

Overall big strides in AP2 highway over the last two updates, equal or better than my best AP1 experience... the biggest room for improvement appears to be on the city streets and recognizing when cars are parked at a light ahead. Very curious how additional cameras and stop light recognition are going to affect things.
 
This is about tomtom and their partnership with Nvidia. Building HD maps and localization. More importantly, being able to process crowdsourced detected changes in realtime and publish new maps in realtime.

Some cool stuff on that here as well:
TomTom HD Map & RoadDNA | TomTom Automotive

As tesla doesn't use lidar, there's an inmovative radar option for localization by tomtom and Bosch
TomTom and Bosch Collaborate on World First: Radar Localisation Map Layer for Autonomous Driving

Also working with Qualcomm
Qualcomm and TomTom Announce Plans to Accelerate the Future of Autonomous Driving (AMS:TOMTOM)
 

Attachments

  • IMG_3185.JPG
    IMG_3185.JPG
    178.5 KB · Views: 64
My guess as to why they switched to Tom Tom: they were told it would start off worse, but get better over time, with full parity coming in 2 months.

TomTom has invested heavily since 2011 for a "realtime mapmaking platform" that can automatically process and validate data from sensors.

But once you have that, you will need as many cars and sensors as possible, to become the winning map.

As such Tesla and TomTom should be a win-win. Tesla has the cars and sensor data. TomTom has the machinery to create valuable data out of it.

Apple and now UBER are also using tomtom data and 80% of carmakers are testing their HD maps. Volvo (a partner of UBER in autonomous driving) is also working with tomtom on autonomous driving.

TomTom key partners are Bosch (a Tesla supplier) , Nvidia and Qualcomm. It's most notable competitors are HERE - now owned by BMW, Audi and Daimler and Mobileye who have partnered with HERE and parted with Tesla. Don't know if I should name Google, because carmakers are reluctant to work with them because of the data snooping. The three German carmakers (consortium) even acquired HERE as a defensive move to steer away from being dependent on Google.

There was also an insightful article in Economist on cartography which also claims Ford, General Motors and Chrysler are working with TomTom on HD maps. They want to steer clear from the German consortium HERE, which is now regarded as a map owned by competitors. This HERE map is also what was featured in the Garmin/Navigon offer that Tesla used to source. So Tesla probably also perceives the HERE map as a German competitor. I've spoken some experts in the field claiming that HERE has much difficulty gaining foothold in the US car industry.